The VW emissions scandal and software

Bernard,

VW software scandal: chief apologises for breaking public trust

Martin Winterkorn orders external investigation after US regulators found cars gave inaccurate data on toxic emissions

Above are the headline and sub-head from this Guardian article.

Seems to me this could be a bug, not a deliberate evasion. If it’s a bug then VW’s engineering processes are broken and the penalties should be even harsher than if VW lied and falsified knowingly. Burden of proof should remain on accuser, who should not impute or assume ill intent on VW’s side.

Wayne

________

From the excellent Vox: Volkswagen’s appalling clean diesel scandal, explained

Updated by on September 21, 2015, 1:40 p.m. ET

 Worth reading for a clear and reliable explanation of what VW did. It was not a bug but a design feature.

Bernard

________
I still want a trial and more direct evidence. It’s all but incomprehensible to me that VW would deliberately, knowingly and with malice aforethought ship cars that behaved like this. It seems almost suicidal. Why would anyone do that? How could thousands of engineers know for years that they were shipping with cheating devices and no one, not one of them, ever once blew a whistle?

Almost certainly it’s a trivial fix in the software to disable the disabler – just sever one path of execution and you’re done.

So I’m not on board, not yet anyway, with tossing VW in the clink.

When I was briefly a military judge I was against convicting lightly but for punishing quite severely upon having convicted. I found the other judges to be quicker to convict, yet once having convicted they were reluctant to punish harshly.

Wayne

__________________

The more articles I read about this, the more I feel that VW admits that its engineers purposely set out to deceive the testing agencies.

The CEO, it seems to me, pretty much admits it.

If it were a glitch I think he’d be speaking about this differently.

Your remarks as a code expert at the highest level, and your views about proper investigations, are valuable.

Volkswagen unveiled the 2016 Passat on Monday at the Duggal Greenhouse in New York. Bryan Thomas for The New York Times

“…  In the lab, the two VWs performed flawlessly. But when they were taken out on the roads in California, they were belching out levels of nitrogen oxide that were 30 to 40 times higher than the regulatory standards. Even the heavy-duty trucks the researchers had tested had never performed that poorly by comparison. … ‘It just didn’t make sense,’ said John German, one of the leaders on the project at the transportation council. ‘That was the real red flag for us.’”

_____________________________

Yes, I think you’re right, it seems to have been deliberate. Barbara says she read or heard that most or all auto companies are doing the same thing to one degree or another, cheating on emissions tests, and all feel they have to do so in order to stay competitive. Still amazes me that it went undetected for so long. Did it never occur to anyone to perform two different kinds of tests, the standard one when stationary and a more realistic underway one, to rule out exactly such shenanigans? Did no engineer in any of these companies ever have moral pangs sharp enough to blow a whistle loud enough for the outside world to hear?

Wayne

_________________________
 This Slate article gives some details and context. An argument for making all code open to investigators
Bernard

______________
This statement in the Slate article is wrong: “The car of the future is a computer with wheels.” It should read, “Today’s cars are tens or hundreds of cooperating computers bundled onto wheels.”

And this Slate sentence is wrong too: “…wherever there’s software, you’ll also find bugs, hacks, and blue screens of death. The stakes are high: If your car crashes, it crashes.” You will not find bugs in the NASA Mars rover. And there’s no need ever to have blue screens of death any more. My mission in life now is to spread the word on this, and to demonstrate in practice in the real world that it’s true.

Wayne
_________________

I encourage you to post your Netherlands talk here, or tell us about it here. While the VW scandal doesn’t seem to be a bug, the software was correctly doing what the engineers and executives intended, the world of software products is filled with flaws and failures at many levels.

How can consumers demand better systems? Isn’t this an example of market failure?

Here is a post from a respected (as far as I know) group blog I look at. I suppose this author is an academic in the social sciences.

Her question, which I think is a good one: How could this not have been widely known within the appropriate parts of the company?

Suppose some small team decided to do this. How would they personally benefit, unless they told the people to whom they reported (and determined their promotions and bonuses)? It seems to me that there must have been some discussion of these plans to meet the design requirements to pass the emissions tests and to achieve certain performance goals. If this originated at the engineering level, then it must have been known some distance up the chain of command. Indeed, the idea may well have originated somewhere in the chain of command, filtering both up and down (but not sideways to the people who designed the tires or chose the paint colors or handled sales, and so on).

Bernard

________

A comment on Waring’s piece, below, sort of captures it. Maybe only a few people higher up knew that the sim/dev code would deliberately be left in after shipment. Everyone else thought it was a test artifact.

casmilus 09.24.15 at 10:29 am

Is it possible this was some sort of simulation/development code that got left in by mistake? No, I don’t think so either, but I wondered if they’d tried that as an excuse.

Maybe it’s the truth, but they’re too embarrassed to admit it, for damage to the image of German technical competence?

Wayne

_______________

A good analytical article about the ways various nations handle auto testing.

And it’s the same as our financial meltdown, by which a firm that intended to issue a bond or some other security would hire one of the ratings agencies to rate the bond. An invitation to corruption on the part of the ratings agencies, similar to the firm that European auto firms hire to run their tests.

Bernard

Leave a comment

Filed under Software

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s